Using MEME-chip for motif discovery in sequences not derived from Chip-seq?

Question

[seta]

Hi everybody,

I use MEME-chip instead of using MEME for motif discovery on DNA sequences due to just large DNA sequence data. Please let me know whether the results OK as the sequences haven't derived from Chip-seq, and also not followed the suggested rules by MEME chip (such as similar sequence length as 100-500bp)? Also, would you anybody please let me know how to get the tabular format results from MEME-chip web too to more easier handle the results? thanks a lot in advance

Answer 1

[Manvendra Singh]

I am quite satisfied with RSAT tools for motif discovery. Its easy to use, quite user friendly, web based and command line. you can manipulate your sequences, analyze them for multi format matrices, motifs, and many. Just explore it here.

Answer 2

[Ian]

I would not worry too much about the 'ChIP' part of the name, as MEME has been optimised to handle large numbers of sequences that is typical of ChIP-seq data. By default MEME-ChIP only uses the top 600 sequences of any submitted set, moreover only the central 100bp of sequence is used to discover motifs.

As for the plain text version of the results when the results are returned you can follow the link that says "The full MEME-ChIP analysis can be downloaded as a gzipped tar file from here."

Within this you can find meme.txt in the meme_out directory. Plus many others.

[seta]

Thanks Ian. Actually, I tried MEME-chip and also RSAT that the other friend suggested for the same data and found that MEME-chip is more faster. I'm a bit in a doubt as the MEME-chip randomly select just up to 600 bp and centralized to 100 bp and then find the motif. So, how it can work well for motif discovery some sequences with 2000 bp or more in length?

[Ian]

I think MEME-ChIP actually just selects the first 600 sequences. I don't think any motif discovery tool is going to work well with long sequences, because the background noise becomes a problem. Which is why the recommendation is short sequences of less than 500bp. However in the day of ChIP-chip I managed to analyse sequences of ~1kb. I would also recommend Weeder2, I have had a lot of good experience with it. I wrote a review for Weeder2 here on Biostar.